Semiconductor processing generally involves subjecting substrates, such as semiconductor wafers, to a series of processes to form layered structures on the substrates. During the course of such processing, it is often desirable to test the outcome of one process conducted on a substrate or batch of substrates prior to conducting another process on the same substrate or batch, in order to determine whether the first process was successful. If it was not successful, the substrates may be discarded or salvaged, if possible, prior to subjecting them to additional processes. Similarly, if it is determined that a process carried out on one batch of substrates was unsuccessful, it is usually desirable to identify and correct the problem prior to carrying out the same process on another batch of substrates so as not to risk having to discard additional substrates.
Various methods have been devised for testing whether certain processes have been successful. For example, in CVD processes, substrates have been weighed both prior to and after processing to determine whether, and by what amount, the weight of the substrates was increased as a result of the deposition process. If the deposition process was successful (i.e., a layer of material having a desired thickness was deposited on the substrates), the amount by which the weight of the substrates increased should fall within a predetermined range. If the change in weight of the substrates falls outside of the predetermined range, it can be assumed that the deposition process was unsuccessful.
One problem with such testing methods is that they are often performed outside of the tool in which the process is carried out on the substrates. The substrates must therefore be removed from the tool and transported to the testing device for testing. This typically requires that the substrates first be loaded into cassettes prior to being transported to the testing device. In the meantime, if processing is continued on another batch of substrates while the previous batch is being tested, there is a risk that additional substrates may have to be discarded when the process results on the previous batch are found to be unacceptable. Alternatively, if processing of the next batch of substrates is discontinued pending the results of the testing, the idle time adversely affects throughput.